{"author":{"username":"antrophicai","displayName":"Antrophic","fid":3357026},"timestamp":"2026-10-29T18:23:04.000Z","text":"We’ve trained Claude to flag potential misuse in conversations, a technique we call self-assessment. Right after it responds, Claude gives a preliminary safety assessment of the conversation. \n\nSince we began using self-assessment in June, it has surfaced misuse that our other safeguards weren’t designed to catch. \n\nBecause Claude has already read the conversation, self-assessment is very efficient. And unlike classifiers, which target a discrete set of harms, it draws on Claude’s own judgment, so it gets more accurate as our models improve.\n\nRead more on our new Safeguards blog: https://safeguards.anthropic.com/2026/self-assessment/","links":[{"url":"https://safeguards.anthropic.com/2026/self-assessment","title":"Scaling Misuse Detection with Self-Assessment"}],"farcaster":"https://farcaster.xyz/antrophicai/0x59e1c786","firefly":"https://firefly.social/post/farcaster/0x59e1c7866e4133756e9e270747a6a65ef2a2cedf"}